Plotting issues in netVisual_circle / netVisual_individual and hierarchy layout (self-loop orientation & tiny nodes)

[songflowerr]

Hi jinworks！
I’m experiencing a couple of plotting issues when using CellChat visualization functions, and I’m not sure whether this is an environment-related bug (e.g., igraph/graphics device) or something in the plotting code. I’d really appreciate your guidance.

1. netVisual_circle() / netVisual_individual() self-loop arcs appear flipped (upside-down)

When I draw the circle plot, the self-loop arcs are rendered in an unexpected orientation (they look “vertically flipped” compared to what I would normally expect). The same issue appears in netVisual_individual().

Code:

P <- netVisual_circle(
cellchat@net$count,
vertex.weight = groupSize, # node size proportional to cell number
weight.scale = TRUE, # scale edge weights to edge width
label.edge = FALSE,
title.name = "Number of interactions"
)

Screenshot:

2. Hierarchy plot: filled/open circles and arrowheads are extremely small (almost a dot)

In the hierarchy plot, the filled circles, open circles, and arrowheads are nearly collapsed into a dot. Adjusting vertex.size.max does not fix it—it only rescales the overall figure rather than making these elements readable.

Screenshot:

Session information

ggplot2: 4.0.2

igraph: 2.2.1

(If you’d like, I can also provide full sessionInfo() output and the CellChat version.)

Questions

Is the self-loop orientation issue a known bug (possibly related to igraph plotting defaults or device coordinates)?

For the hierarchy plot, is there a parameter or recommended approach to control the size of the filled/open circles and arrowheads independently (not just global scaling)?

If this is version-specific, are there known compatible versions of igraph / ggplot2 that avoid these issues?

Thank you very much for your time and for maintaining CellChat.

Best regards,
Songflower

[sqjin]

@songflowerr Please check the session infomation of our tutorial, and try to use the same version of igraph and ggplot2